breeze
The open-source IT platform that comes with the workers. RMM + PSA in one system, with a governed AI operator built in.
Breeze bundles RMM and PSA with a built-in AI operator
An open source IT platform for MSPs and internal teams that monitors, patches, and resolves tickets with agent assistance.
What Breeze provides
Breeze is a full IT platform for managed service providers and internal IT teams. The README describes monitoring, patching, remote access, scripting, and network discovery, combined in the same system with ticketing, quotes, billing, invoicing, and a service catalog. The distinguishing claim is that the AI is built into the core rather than added later: an AI operator uses the platform on the operator's behalf, investigating alerts, remediating issues, resolving tickets, and documenting what it did, and it only interrupts a human when a decision is needed. The project is free and open source under AGPL-3.0 and is designed to be self hosted or cloud hosted. The README contrasts Breeze with other RMM tools that hand over a dashboard and leave the labor to the operator, arguing that Breeze instead hands over a team. It also emphasizes that RMM and PSA live in one system, so the ticket, the time entry, the quote, and the invoice stay where the work happens instead of across separate products. Self hosting is described as never metered, with unlimited devices and every module in every plan, and the agent is a single Go binary that is cross platform with a small footprint. The README presents unlimited self hosted devices and a single binary agent as the practical advantages for a team that wants to avoid per endpoint pricing from commercial tools.
Feature areas and security
The feature list is broad. Device management covers hardware and software inventory, real time health with configurable thresholds, hierarchical configuration policies, advanced filtering, network discovery through ARP, ICMP, port, and SNMP scans, custom fields, and configuration drift tracking with CIS checks. Remote access includes a terminal, a file browser, remote desktop with multi display and clipboard sync, native Tauri viewer apps, activity monitoring, and a built in TURN relay for WebRTC. Automation covers remote scripting in PowerShell, Bash, and Python, patch management with maintenance windows, alerts, playbooks, deployments, and a self healing watchdog. The PSA side adds tickets with SLA tracking, quotes from a catalog, billing where time and parts become the invoice, a service catalog, and a customer portal. Backup covers Restic snapshots to S3, bare metal recovery for Windows, Hyper-V and SQL agents, and M365 cloud to cloud backup. Security is detailed in a table: Argon2id passwords, JWT with short expiry, TOTP MFA, AES 256 GCM at rest, TLS 1.2 or higher, RBAC with PostgreSQL row level security, Redis rate limiting that fails closed, Zod input validation, and five CI scanners including CodeQL, Gitleaks, npm audit, govulncheck, and Trivy. The detailed security table signals that Breeze treats its privileged access seriously, and the fail closed rate limiting shows the design prefers denial of service over leaking an auth endpoint.
AI operator, integrations, and deployment
The AI operator is included free and appears as a context aware chat on every page, equipped with tools to query devices, run diagnostics, and execute remediations. Every AI action is checked against a risk engine before it runs, impactful actions require human approval, and critical operations are blocked entirely. Users bring their own Anthropic API key, and external agents such as Claude.ai, ChatGPT, and Cursor can connect through a built in MCP server using OAuth 2.1 with PKCE, which the README calls the first MCP server shipped in an RMM. Integrations also cover EDR products SentinelOne and Huntress with risk classified actions. Deployment offers a cloud hosted option on breezermm.com with US and EU regions in public beta, or a self hosted guided setup using Docker and Docker Compose that generates an env file and a compose file in an empty directory. The README also describes a managed AI ops offering where a team of AI agents works the queue inside the customer's own Breeze instance while the vendor reviews resolutions and tunes the agents, with the risk engine holding approvals throughout. A security whitepaper maps the practices to SOC 2, and vulnerabilities are reported to a dedicated security address.
Editorial conclusion
The platform is published under the AGPL-3.0 license and its most recent commit was on 2026-08-26.
Community notes